3 *) [WIN] Use a permanent pool for allocating the SystemRoot
4 environment variable. Sakamoto [ringring@zb4.so-net.ne.jp]
6 *) [WIN] Fix starting of scripts under AP2.
8 *) Do the connect() to the application *after* collecting a chunk
9 of client data. This reinstates the pre-2.4.0 behaviour.
10 James Jurach [muaddib@erf.net].
12 *) Remove an assert that was triggering on WIN when spawn() failed.
14 *) Provide the NO_SUEXEC_FOR_AP_USER_N_GROUP macro for building
15 mod_fastcgi with the AP13 suexec behaviour (don't use suexec
16 if httpd's user and group match that needed for the application).
18 *) Prevent the use of all but the "auth" directives from being used
19 anywhere but in global scope. Prevent more than one instance
20 of the FastCgiWrapper directive.
22 *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of
23 INTERNAL_SERVER_ERROR (500) when there are configuration issues
24 or the script does't exist. Suggested by Jeff Lawson [bovine@ud.com].
28 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
29 and a vhost configured to use the same uid/gid as the main
30 server, mod_fastcgi would not bother using the wrapper (suexec)
31 because its effective uid/gid was already appropriate. This is
32 consistent with Apache's v1.3 mod_cgi behaviour. There are two
33 problems with this approach: 1) when FastCgWrapper is in use
34 mod_fastcgi's process manager keeps its root privileges (as its
35 real uid/gid) so it can terminate the applications its starts -
36 this privilege was being passed to applications when the use of
37 the wrapper was bypassed 2) wrappers are often employed to
38 perform functionality beyond setting the uid/gid - by not calling
39 the wrapper under certain circumstances, application invocation
40 environments were inconsistent. With this change, the wrapper is
41 always used (when enabled) under both Apache 1.3 and 2.
42 Reported by Michael Richards [michael@fastmail.ca].
44 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
45 for dynamic application invocation when the FastCgiWrapper is in use.
46 Reported by Michael Richards [michael@fastmail.ca]
48 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
49 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
51 *) Eliminate the logging of "incomplete headers (0 bytes) received from
52 server" when a client aborts.
54 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
55 servers. Philip Gladstone [philip@okena.com]
57 *) [*nix/AP2] Prevent the module from being initalized twice at startup
58 (resulting in confusing error messages to the log).
60 *) Eliminate the need for SetHandler or AddHandler with static or
61 external applications.
63 *) Limit PM requests to start a dynamic application
64 to 5sec to prevent endless spinning (this is a drop-dead
65 limit that should only occur if the socket/named_pipe directory
66 is removed out from under a running server).
68 *) [*nix] Change the default socket directory from /tmp/fcgi to:
70 Apache2 - RUNTIMEDIR/fastcgi
72 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
73 for use with wrappers (in lieu of finding the user/group associated
74 with a virtual host - under Apache2 this isn't accomodated).
76 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
77 apr_proc_create() changes.
79 *) Log when invoking and restoring the restart backoff policy.
81 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
82 OVERLAPPED structure).
84 *) Eliminate need for dummy files for external servers under Apache2
86 *) Fix auth compatibility mode handling for access checker and authorizer
88 *) Fix HEAD request handling. Based on a patch by
89 Chris Lightfoot [chris@ex-parrot.com]
91 *) [*nix] When autoupdate is enabled touch the socket when restarting
92 the processes to prevent further requests.
93 Eckebrecht von Pappenheim [evp@heise.de]
95 *) Apache 2.0 support.
97 *) [WIN32] Don't read from a potentially closed named pipe.
98 Philip Gladstone [philip@okena.com]
100 *) Require the Apache version 1.3.6 or later to eliminate some signal
103 *) [WIN32] Use asyncronous io with named pipes instead of polled
104 nonblocking io. This should eliminate the last of the npipe issues.
106 *) Handle an application returning a complete and valid response without
107 having consumed all of the data sent to it.
109 *) Consume remaining client data (RESPONDERs only) if any.
111 *) Add support for backing off attempts to start applications that continuously
112 fail to start. Three new macros defined in mod_fastcgi.h control this
113 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
115 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
116 applications that do not (properly) support the shutdown event (this
117 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
118 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
119 interval between signaling a proper shutdown and wacking the process(s)
120 with a TerminateProcess().
122 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
123 setting it was just plain broken.
125 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
127 *) Fix 'FastCgiConfig -autoUpdate'.
129 *) Fix 'FastCgiConfig -flush'.
131 *) Prevent silly maxProcesses and processSlack combinations.
132 Dmitry Dorofeev [dima@yasp.com]
134 *) Properly handle the killing of idle processes when one takes a long time
135 to exit once signaled down (or the config is funky).
136 Dmitry Dorofeev [dima@yasp.com]
138 *) Always kill the youngest instance of an application. Suggested by
139 Dmitry Dorofeev [dima@yasp.com]
143 *) Delay the logging of write errors to the pm to account for shutdown/restart.
145 *) (Win32) An assortment of fixes.
147 *) Fix some broken casts that were likely the cause of an assert.
149 *) Win32. Eliminate forward slashes from the named pipe path name.
150 Gerald Richter [richter@ecos.de]
152 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
153 applications it spawns. [ryans@amazon.com]
155 *) Added support for the -flush argument to FastCgiConfig.
156 Eric Sit [esit@alum.mit.edu]
158 *) Change the "which call to module_init() is this" check to a more
159 reliable approach. Doru Petrescu [pdoru@kappa.ro]
161 *) Close the old pipe file descriptor in apache main on USR1/HUP
162 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
164 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
165 [tetsuya@secom-sis.co.jp]
167 *) Eliminate the use of locks to assist in the clean shutdown of
168 applications. Instead, it is assumed that applications handle
169 termination signals properly (this is now embedded in the C
172 *) Fix Win32 process termination. Proper operation requires the use of
173 an updated application lib (termination is now signalled with an
174 Event and handled by specialized thread).
176 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
178 *) Added code so if the last instance of a dynamic application died without
179 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
180 configuration allows the last instance to be killed, then allow it to die).
181 Andrew Benham [adsb@bigfoot.com]
183 *) Fix the loadFactor calculation used to determine when dyanmic
184 applications could be killed off due to low demand [adsb@bigfoot.com].
186 *) Fix a deadlock condition that could occur with Win32 named
189 *) Fix a potential deadlock condition when FastCGI application
190 sent responses while still reading the client request (POST data).
194 *) Allow absolute pathnames in the -socket argument. Suggested by
195 Christian Jaeger [christian.jaeger@sl.ethz.ch].
197 *) Don't invoke suexec when the user/group for the fastcgi application
198 is the same as the apache main server. This is consistent with
199 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
201 *) Reset the apache drop dead timer upon successful read or writes
202 to/from the client. This eliminates timeouts that were occuring
203 during the large file transfers to/from slow clients.
205 *) Support generic wrappers such as cgiwrap by eliminating dependencies
206 on Apache's SUEXEC, renaming the FastCgiSuexec directive
207 FastCgiWrapper and eliminating any checks regarding the target
208 application (this is the repsonibility of the wrapper).
210 *) Fix a nasty bug that occurred when a client aborted a POST request
211 before the connection to a dynamic FastCGI application was opened.
212 The application's lock file descriptor wasn't setup but was being
213 closed which resulted in FD0 being closed. Normally this is open to
214 /dev/null and should pose no problems except that because the FD
215 was available it was being returned by Apache's accept(). This
216 caused it to be registered for pool cleanup. mod_cgi though moves
217 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
218 cleanup. Problem identified by checksum@163.net.
220 Changes with mod_fastcgi 2.2.8
222 *) Eliminate the concept of disabled applications. If a failure
223 occurs trying to setup an application (e.g. bind() error) its
224 tried repeatedly every init-start-delay seconds.
226 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
228 Changes with mod_fastcgi 2.2.6
230 *) Shutdown the PM when Apache appears to have disappeared.
232 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
234 *) (Win32) More dynamic fixes.
236 *) (Win32) Eliminate the per application named pipe mutex. Its now
237 per process to keep the lib happy (the use _FCGI_MUTEX_ should
238 removed from the lib). This allows pipe-based applications to
239 now handle multiple simoultanous requests (one per process).
241 *) Increase the number of open FDs we look to close when spawning apps.
243 *) Prevent an assert from popping unnecessarily.
245 *) (Win32) Add support for interpretter scripts.
247 *) (Win32) Fix named pipe handling (problems with large responses).
249 *) (Win32) Remove the "can exec" check.
251 Changes with mod_fastcgi 2.2.4
253 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
254 Rob Saccoccio [robs@chelsea.net]
256 *) Remove request type restriction (GET and POST only) in order to support
257 Web DAV requests. Scott Robertson [sroberts@codeit.com]
259 *) Allow requests for URLs such as /server/some/other/qualifier to match a
260 FastCGI server defined as /server. This was done primarily for Java
261 Servlets, but is generically useful.
263 *) Change the comm between the PM and the request handlers from a regular
264 file to a pipe and an assortment of other dynamic fixes.
266 *) Log dynamic process termination scheduling and the resulting
267 process exit notification.
269 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
270 application process from being killed off due to low demand.
272 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
274 *) Fix bugs that prevented dynamic processes from being shutdown when
275 the load subsided. Lars Heete [heete@do.isst.fhg.de]
277 *) Prevent dynamic processes from being scheduled to be started before
278 the init-start-delay or restart-delay period expires. This prevents
279 the queuing of process start requests while an application which has
280 a long initialization period starts up.
282 *) When suexec is enabled, allow the process manager to become root again in
283 order to signal applications it spawned.
285 *) Change the "server started" log message level from INFO to WARN.
287 *) Fix dynamic server "Connection Refused" handling.
289 *) Fix demand-based dynamic process spawing (uses a more portable approach).
291 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
292 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
293 longer than this period. It applies to the initiation of connections as
294 well (and thus is similar to appConnTimeout). Default is 30 seconds.
296 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
297 is more platform portable/predicable.
299 *) Leave STDOUT and STDERR open to the main server error_log. This should
300 eliminate a pile of protocol errors and having to track down 3rd party
301 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
302 be fixed in the application to use the FastCGI I/O, it just allows these
303 applications to run when previously they'd crash and burn.
305 *) Add the pass-header arg to server directives.
307 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
308 is done for CGI) - at least until PR2580 is resolved.
310 *) Initialize the default (empty) environment in a more socially acceptable
313 *) Miscellaneous doc improvements based on notes I've collected up over the
316 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
317 that resulted in not or improperly terminated strings.
319 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
320 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
322 Changes with mod_fastcgi 2.2.2
324 *) Added support for blocking connect()s by setting appConnTimeout to 0.
325 Non-blocking connect()s (the default) can be troublesome on some platforms.
326 Its expected that blocking connect()s will become the default (and
327 non-blocking connect()s will become optional) in the next release.
329 *) Dump a compile time error if the version of Apache is too old.
331 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
332 to Apache releases prior to 1.3.6.
334 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
338 Always restart a failed dynamic application if it is the last instance.
339 This means there once started there will always be at least one process
340 instance of a dynamic application.
342 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
343 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
344 the PM to start (another) application instance may help empty it faster.
346 Change two sleep() calls to select() based snoozes because alarm() is in
347 effect and sleep() and alarm() don't always play nice together.
349 Fixed a couple of error messages.
351 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
352 was ignored and the default was always used.
354 *) Allow the -initial-env argument to be used to pass variables from the
355 Apache process environment to the FastCGI server (by specifying a
356 variable name without the "=" or a value). [suggested by
357 Martin Lichtin <lichtin@oanda.com>]
359 *) Cleanup some debug macros.
361 *) Improved script stderr handling. [based on suggestions by
362 David Birnbaum <davidb@chelsea.net>]
364 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
365 updates to the INSTALL doc.
367 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
368 5 to 100. Its still configurable with the -listen-queue-depth option.
369 This should help eliminate the FAQ - Why do I see "Connection Refused"
372 *) Fix a bug in FastCgiExternalServer that broke support for external servers
373 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
375 Changes with mod_fastcgi 2.2.1
377 *) Updates to the INSTALL doc to describe building as a DSO.
379 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
380 Apache was running ("apachectl restart" does this), the contents of the
381 dynamic directory (dynamic sockets and the mbox) would be blown away
384 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
385 Apache header file) to prevent a compile error on systems without
388 *) During auth requests, the subprocess_env table was left holding variables
389 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
390 were passed to other processing phases (such as CGI/FastCGI). The
391 subprocess_env is now restored to its pre-auth condition.
393 *) Added a FastCGI Authorizer Role compatibility mode which implements the
394 specification to the tee. Use -compat arg with any of the Auth server
395 directives. This is intended for new and existing FastCGI authorizer
396 applications that require compatibility with other server implementations.
398 *) Fix logging from the process manager. Previously, all calls to the
399 logging routines used NULL for the server_rec. This works fine in a dev
400 environment, but results in most of these messages being tossed by
401 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
403 *) Always setup fr->header so that in the event the FastCGI server
404 doesn't get a valid FastCGI protocol header over the wire, we
405 can still print our error message to the log from do_work().
407 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
408 defined (Don Locrasto [locrasto@iceminer.com]).
410 Changes with mod_fastcgi 2.2.0
412 *) Serious rewrite of mod_fastcgi.html. An example conf will have
413 to wait until another time - I'm burned out on docs updates.
415 *) Add to the mod_fastcgi to the server version string.
417 *) SCO doesn't like the const in the arg to inet_addr() and
420 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
422 *) Deleted redundant header files from fcgi.h.
424 *) Full path names are no longer required for directives which take
425 file names as arguments. Like other Apache directives you can
426 specify paths from server_root (don't start with a "/").
428 *) The fcgiKillMgr is gone. This intermediate process sat between
429 the Apache parent and the FastCGI Process Manager under Apache
430 1.2. The process manager tries to name itself fcgi_pm (it used
433 *) Authentication, Authorization, and Access phases are now
434 supported. I've bent the FastCgi spec a bit: as many of the standard
435 environment variables (that were easy) are sent (the spec says don't
436 send some by name), all headers (except Status) sent by the auth
437 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
438 as environment variables rather than just those prefixed by
439 "Variable-". Custom responses for auth failures aren't supported
440 (yet.. we'll see what the demand for this is like). In addition to
441 the FastCGI protocol defined environment variable "FCGI_ROLE" being
442 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
443 set indicating which of the three phases is being processed. See
444 the docs for info on the new per-directory directives:
446 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
447 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
448 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
450 *) The code has been broken into logical chunks making figuring it
451 out much easier. All of the #defines a user may want to change are
452 now in mod_fastcgi.h.
454 *) All the logging has been converted to the "supported" Apache
455 logging routines, ap_log_error() and ap_log_rerror(). Log entries
456 for request specific errors should now be directed to the correct
457 server errorlog (if your running more than one). Some attempt has
458 been made at setting appropriate log levels (comments of course are
459 welcome) and printing errno info when its of value. All of log
460 messages now have "FastCGI" in them (nice for grep).
462 *) All of the calls to Apache routines have been updated to the ap_
463 convention, eliminating the need for the compat header.
465 *) By default we no longer flush every piece of data we get from the
466 FastCGI server to the client. This allows the FastCGI server to
467 release without having to wait for the client flush to complete.
468 The old behaviour can revived with the -flush argument to AppClass.
470 *) I renamed some of the directives for consistency (the old names
473 AppClass -> FastCgiServer,
474 FCGIConfig -> FastCgiConfig,
475 ExternalAppClass -> FastCgiExternalServer
477 *) Directive arguments are no longer case sensitive.
479 *) The module now uses hard_timeout() rather than soft_timeout. This
480 means that the module will longjump out of the request if the
481 drop-dead timeout expires (set with Apache's Timeout directive, the
482 default is 5min) or if the client closes the connection (SIGPIPE).
483 This is more typical for a Apache modules and I'm not convinced we
484 properly handle all of the error cases well enough to use
485 soft_timeout (I can think of one place we don't anyway). This means
486 your FastCGI application can see SIGPIPE.
488 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
489 Apache. Without this the default behaviour is exit().
491 *) It should (I think) run now under DSO now.
493 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
495 *) The broken -initial-env bug ins 2.1b1 is fixed.
497 *) Maybe some other stuff.
501 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
502 commit notes and stick 'em in here.. when I get some free time. ;)
507 *** Originally from docs/README..
509 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
510 From: Stanley Gambarin <stanleyg@cs.bu.edu>
512 *) Yet more changes to the source distribution. Separated out
513 the Tcl dynamic string and buffer libraries into separate files.
514 Did the same for the OS library, however, it is not really an
515 abstraction... more like a bunch of wrappers. Added to DEVNOTES
516 a note about the problem of Apache not allowing including header
519 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
520 From: Stanley Gambarin <stanleyg@cs.bu.edu>
522 *) More changes to the installation scripts. Theoretically we
523 should be able to to use the same Makefile/installation script
524 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
525 layer to use the same source code for both 1.2.x and 1.3.x sources.
527 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
528 From: Stanley Gambarin <stanleyg@cs.bu.edu>
530 *) Created Makefile to compile the module into the archive library,
531 just like the proxy module. Also initial draft of the installation
532 script, which should work for both 1.2 and 1.3 sources of Apache.
534 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
535 From: Stanley Gambarin <stanleyg@cs.bu.edu>
537 *) More source reorg. Separated out some header files. Need much
538 more work on this. Updated the TODO file. Added fcgivers.h and
539 CHANGES file to track the history.
541 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
542 From: Stanley Gambarin <stanleyg@cs.bu.edu>
544 *) Source reorganization. This is done to provide a basis for later
545 functionality implementation. See TODO file for more information
546 of what possible future enhancements are possible/desired.
548 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
549 From: David MacKenzie <djm@va.pubnix.com>
551 *) Create the "dynamic" dir and "mbox" as the user specified in the
552 User and Group directives instead of as root. They are created with
553 restrictive permissions, and then the module checks to see whether the
554 user specified in the User and Group directives has read, write, and
555 execute permissions on them. Those permissions had been explicitly
556 denied by creating them as "root".
558 *) Some documentation was garbled or incomplete.
560 *) When a FastCGI application can't be execl'd, the code used a
561 value of errno that may have been stomped on by intervening system
562 calls. That happened on BSD/OS 2.1, where the error_log was
563 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
564 An intervening stdio call was setting errno as a side-effect, which
565 it has the right to do.
567 *) Erroneous fprintf arguments, a missing return value, an unused
568 function, missing declarations, and other problems detected by gcc
571 *) Duplicated code to create the lock file name merged into a single
572 function, closing several memory leaks.
574 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
575 FCGI app in a given dynamic class would live forever, even if it
576 hadn't been requested for weeks.
578 *) The test for whether to keep looking for dynamic app victims to
579 kill was backwards, so no dynamic apps were ever selected as
582 *) The killInterval and updateInterval were being ignored; the
583 sigsuspend() forced recalculations only when a child died,
584 instead of at the intervals specified.
586 *) When a dynamic app couldn't be started, the program SEGV'd when
587 trying to free the ipcAddrPtr twice.
589 *) Remove the dead lock file and socket when cleaning up after a
590 server whose last child has been killed, so the FastCGIHandler
591 isn't fooled into thinking there is still a process serving
594 *) Make file locking robust in the presence of signals.
596 *) Rename some badly named variables.
598 *) Fix typos in many comments.
600 *) Fix some memory and file descriptor leaks.
602 *) Make the blocking kill of a server closer to working.
604 *) More fixes to calculations in dynamic application management.
607 *** Originally from docs/README.OMI..
609 What's New: Version 2.0b1, 16 Apr 1997
611 *) Implemented a mechanism by which FastCGI applications are started by
612 the web server on the first request and continue running. Also, a
613 heuristics have been implemented to allow for dynamic killing of the
614 running FastCGI apps. The configuration options are supplied via
615 the new FCGIConfig directive, that is described in the
618 *) When performing internal redirect, since the original request's body
619 has already been read, do not allow the redirected request to think
620 that is has one [body].
622 *) More conditional compilation for OS/2.
624 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
626 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
628 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
629 force some output to be written in the period of inactivity. This
630 circumvents the Apache's buffering during the server-push.
632 *) Added a header-parser function placeholder for compatibility with an
635 *) Additional checks have been placed with regard to permissions of the
636 FastCGI processes. These include the existence of the file,
637 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
639 *) Implemented a restart cleanup, so that no parentlesses FastCGI
640 applications are left after Apache is restarted and/or terminated.
641 This was accomplished via two independent mechanisms, where the
642 first one prevents a start of the process manager on the first
643 reading of the configuration files. The second mechanism deals away
644 from Apache's implementation of fork() (via spawn_child()) and
645 implements its own forking, thus removing any dependencies on the
646 Apache to cleanup processes during termination phase.
648 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
649 itself in the file conf.h
651 What's New: Version 1.4.3, 15 Jan 1997
653 *) Fixed compilation warnings for various platforms, as well as
654 conditional compilation for OS/2.
656 What's New: Version 1.4.2, 12 Dec 1996
658 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
659 will proceed under this version of Apache web server.
661 *) As the result of porting, the "include virtual" construct of SSI
662 will now work correctly using either <Location> or <Directory>
665 What's New: Version 1.4.1, 4 Dec 1996
667 *) Checks have been removed from the ScanCGIHeaders that provided
668 for the presence of both Status and Location headers as being
669 an error. Contradictory to CGI/1.1 Internet Draft, both of
670 these headers are used by the current CGI applications.
672 What's New: Version 1.4, 22 Nov 1996
674 *) Added the -port option to AppClass, allowing TCP/IP communication.
675 Added the -socket option to AppClass, allowing Unix domain
676 communication via a configurable pathname.
678 *) Added the ExternalAppClass directive, allowing TCP/IP
679 communication with remote FastCGI applications.
681 *) The handler had its own code for generating HTTP response headers;
682 now it uses Apache's. This reduces the size of the module.
683 More importantly, it fixes the bug in which "include virtual"
684 sees the HTTP response headers.
686 *) The response header parser performed very little checking.
687 Now the parser enforces the guidelines in the CGI/1.1
688 Internet-Draft: Status and Location are mutually exclusive,
689 Location can only be a response to GET or HEAD,
690 CGI response headers can't be repeated, etc. (The CGI response
691 headers are the ones the handler interprets: Status, Location,
694 *) The response header parser used to miss CGI headers
695 with no whitespace after the colon, e.g. "Status:200 OK".
697 *) The response header parser sometimes interpreted the first line
698 of content as an RFC822 continuation line.
700 *) The handler implemented a nonstandard version of Location
701 which never used internal redirects. The handler
702 now attempts to implement Location as specified in the
703 CGI/1.1 Internet-Draft. The module documentation explains
706 *) Error log entries from the response header parser were
707 pretty uninformative; they are better now. If a header is
708 malformed, the log entry includes it. If the headers are
709 unterminated rather than malformed, the log entry says that,
710 and says how many bytes were received from the app.
712 *) When the application manager forked a new process, and that
713 process ran into trouble before executing the first instruction
714 of application code, the process used to exit with status =
715 errno. This made certain configuration problems (e.g. incorrect
716 file permissions when server parent is root) quite difficult
717 to pin down. Now the failing child process opens up the error log
718 and writes an informative entry before exiting.
720 *) The module now correctly handles a slash at the end of the
721 DocumentRoot directive. This was a one-line fix.
723 What's New: Version 1.3.3, 17 Oct 1996
725 *) The module now registers its request handler under the name
726 fastcgi-script in addition to the name application/x-httpd-fcgi.
727 This was a one-line addition to the module, but had quite
728 a large impact on the documentation and sample configuration.
730 What's New: Version 1.3.2, 27 Sept 1996
732 *) On some systems (SunOS, Linux?), fopen for append has a bug
733 that strikes when two processes append to the same file.
734 This bug causes the process manager to corrupt the error log.
735 Work around the bug by calling open, then fdopen. A patch
736 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
737 (Reported by Bob Ramstad.)
739 What's New: Version 1.3.1, 17 Sept 1996
741 *) On Linux and SunOS, the Apache default user_id (-1) is not a
742 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
743 Scott Langley, others.)
745 *) On some systems (Linux, some Solaris?, Irix?), connect requires
746 write access to a Unix Domain socket, so provide it. (Reported
747 by Scott Langley, freeform@wired.com, others.)
749 *) If you hit a .fcg file, but there's no AppClass defined, the
750 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
753 What's New: Version 1.3, 4 Sept 1996
755 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
757 *) Module violated the Apache buff abstraction.
758 Fixed. Might now work with SSL (not tested).
759 As part of the fix, eliminated support for
760 Apache 1.0x versions.
762 *) Module failed to chmod the Unix domain listening sockets
763 it created, so protections were set according to the current
766 *) Module forked too many process manager processes: One
767 per AppClass. When the Apache parent ran as root, these
768 process manager processes ran as root. New process
769 manager is a single process (only started if AppClass is
770 used) and runs with same privileges as other children.
771 The new process manager doesn't do any polling, so
772 there's less system overhead than before.
774 *) AppClass insisted on getting at least two arguments. Fixed.
776 *) Module used setjmp/longjmp, causing compiler warnings
777 on some platforms (e.g. Linux.) The module no longer
780 *) Module created listening sockets in /tmp, where they were
781 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
782 directive to give control over the location of listening
785 *) Module wrote error log entries without a timestamp. Fixed.
787 *) AppClass directive wrote error messages to stderr in addition
788 to returning a char * message to the Apache core. Fixed.
790 What's New: Version 1.2, 3 June 1996
792 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
795 Add version string: APACHEVERSION macro to mod_fastcgi.c
797 *) chown FastCGI socket to user_id and setuid to user_id for app
800 *) Modify GetFromStream() by having it call an OS dependent
801 function GetStreamSize(FILE *) which uses FILE internal data member.
802 Linux users might need to modify GetStreamSize().
804 What's New: Version 1.1, 10 May 1996
806 *) If you specify a non-existent executable in the AppClass directive,
807 or if the file exists and it does not have execute permission you
808 get a constant stream of error messages telling you that "program
809 terminated due to a signal".
811 *) The mod_fastcgi module should use the standard Apache error logging
812 facility instead of writing to stderr.
814 Version 1.0, 30 April 1996